I have configured Ethernet Bonding for our RHEL 6.2 Servers. Bonding is working fine but few confusion for me when I configured MRTG to monitor the bonded interface.

MRTG is showing MaxSpeed of the Bonded Interface to be 1250.0 KBytes/s which comes to be 10 Mbits/s. The MaxSpeed for the Slave interface speed (eth0,eth1) looks fine with 125.0 MBytes/s.

My confusion here is since MRTG is showing MaxSpeed of the bonded interface to be 1250.0 KBytes/s, will it be able to handle traffic beyond 10MBits/s or it is MRTG configuration issue only?

Also, would like to know when monitoring a bonded interface, is it OK to monitor the bonded interface only or need to monitor Slave interface also. Any best practices?

Please find the bonding config pasted below.

$ vi /etc/sysconfig/network-scripts/ifcfg-bond0

DEVICE=bond0
ONBOOT=yes
TYPE=Ethernet
BOOTPROTO=static
IPADDR=192.168.11.149
NETMASK=255.255.255.192
GATEWAY=192.168.11.130
USERCTL=no
BONDING_OPTS="mode=1 miimon=100"
$ cat /etc/sysconfig/network-scripts/ifcfg-eth0

DEVICE=eth0
ONBOOT=yes
BOOTPROTO=none
SLAVE=yes
MASTER=bond0
TYPE=Ethernet
IPV6INIT=no
USERCTL=no
NM_CONTROLLED=no

Please clarify.
If any other information is required, please let me know.

Thanks in Advance

You can use iperf tool to test bandwidth of bonding interface.

Installed Iperf and tried pumping data at 40Mbits/sec . MRTG is reporting only 4Mbits/sec. But if I send data at 4Mbits/sec or 7Mbits/sec ,MRTG is reporting fine. So , looks like bonded interface cannot transmit data beyond 10Mbits/sec as MRTG is reporting Max Speed of Bonded Interface to be 10 Mbits/sec. Any suggestion , how to increase the speed of bonded Interface.

Here is Iperf Output for 40Mbits/sec


Server:


======
$ bin/iperf -s -u
------------------------------------------------------------
Server listening on UDP port 5001
Receiving 1470 byte datagrams
UDP buffer size: 122 KByte (default)
------------------------------------------------------------


[ 3] local 192.168.2.149 port 5001 connected with 192.168.2.161 port 51865
------------------------------------------------------------
Client connecting to 192.168.2.161, UDP port 8908
Sending 1470 byte datagrams
UDP buffer size: 122 KByte (default)
------------------------------------------------------------
[ 5] local 192.168.2.149 port 47635 connected with 192.168.2.161 port 8908
[ ID] Interval Transfer Bandwidth Jitter Lost/Total Datagrams
[ 3] 0.0-600.0 sec 2.93 GBytes 42.0 Mbits/sec 0.012 ms 0/2142858 (0%)
[ 3] 0.0-600.0 sec 1 datagrams received out-of-order
[ 5] 0.0-600.0 sec 2.93 GBytes 42.0 Mbits/sec
[ 5] Sent 2142858 datagrams
[ 5] Server Report:
[ 5] 0.0-600.0 sec 2.93 GBytes 42.0 Mbits/sec 0.018 ms 0/2142857 (0%)
[ 5] 0.0-600.0 sec 1 datagrams received out-of-order


Client:

$ bin/iperf -c 192.168.2.149 -b 40Mbit/sec -t 600 -i 60 -d -L 8908
WARNING: option -b implies udp testing
------------------------------------------------------------
Server listening on UDP port 8908
Receiving 1470 byte datagrams
UDP buffer size: 126 KByte (default)
------------------------------------------------------------
------------------------------------------------------------
Client connecting to 192.168.2.149, UDP port 5001
Sending 1470 byte datagrams
UDP buffer size: 126 KByte (default)
------------------------------------------------------------
[ 4] local 192.168.2.161 port 51865 connected with 192.168.2.149 port 5001
[ 3] local 192.168.2.161 port 8908 connected with 192.168.2.149 port 47635


[ ID] Interval Transfer Bandwidth
[ 4] 0.0-60.0 sec 300 MBytes 42.0 Mbits/sec
[ 3] 0.0-60.0 sec 300 MBytes 42.0 Mbits/sec 0.018 ms 0/214285 (0%)
[ 4] 60.0-120.0 sec 300 MBytes 42.0 Mbits/sec
[ 3] 60.0-120.0 sec 300 MBytes 42.0 Mbits/sec 0.013 ms 0/214286 (0%)
[ 4] 120.0-180.0 sec 300 MBytes 42.0 Mbits/sec
[ 3] 120.0-180.0 sec 300 MBytes 42.0 Mbits/sec 0.015 ms 0/214286 (0%)
[ 4] 180.0-240.0 sec 300 MBytes 42.0 Mbits/sec
[ 3] 180.0-240.0 sec 300 MBytes 42.0 Mbits/sec 0.012 ms 0/214285 (0%)
[ 4] 240.0-300.0 sec 300 MBytes 42.0 Mbits/sec
[ 3] 240.0-300.0 sec 300 MBytes 42.0 Mbits/sec 0.014 ms 0/214286 (0%)
[ 4] 300.0-360.0 sec 300 MBytes 42.0 Mbits/sec
[ 3] 300.0-360.0 sec 300 MBytes 42.0 Mbits/sec 0.013 ms 0/214286 (0%)
[ 4] 360.0-420.0 sec 300 MBytes 42.0 Mbits/sec
[ 3] 360.0-420.0 sec 300 MBytes 42.0 Mbits/sec 0.017 ms 0/214285 (0%)
[ 4] 420.0-480.0 sec 300 MBytes 42.0 Mbits/sec
[ 3] 420.0-480.0 sec 300 MBytes 42.0 Mbits/sec 0.020 ms 0/214286 (0%)
[ 4] 480.0-540.0 sec 300 MBytes 42.0 Mbits/sec
[ 3] 480.0-540.0 sec 300 MBytes 42.0 Mbits/sec 0.014 ms 0/214286 (0%)
[ 4] 540.0-600.0 sec 300 MBytes 42.0 Mbits/sec
[ 4] 0.0-600.0 sec 2.93 GBytes 42.0 Mbits/sec
[ 4] Sent 2142859 datagrams
[ 3] 540.0-600.0 sec 300 MBytes 42.0 Mbits/sec 0.019 ms 0/214285 (0%)
[ 3] 0.0-600.0 sec 2.93 GBytes 42.0 Mbits/sec 0.018 ms 0/2142857 (0%)
[ 3] 0.0-600.0 sec 1 datagrams received out-of-order
[ 4] Server Report:
[ 4] 0.0-600.0 sec 2.93 GBytes 42.0 Mbits/sec 0.012 ms 0/2142858 (0%)
[ 4] 0.0-600.0 sec 1 datagrams received out-of-order

The bond interface bandwidth or speed is large than 40 Mbits/sec because iperf report 40 Mbits/sec traffic can transmit on the bond interface. The MRTG reporting is wrong.

Thanks Nini . Let me look at the configuration of MRTG.

To answer your Topic question, what i'd do is identify underlying interfaces speed first (with ethtool for example), and based on your bonding setup figure out the theoritical speed.
Confirm it by using iPerf as mentioned above. Note: for high performance NIC (10Gbps), set the MTU on both sides (client and server) to Jumbo Frame (e.g. 9000) before running iPerf, if you want to max up the bandwidth :-)